1. This study used 4 wheat cultivars (Brigadier, Chaucer, Consort, Reaper) from three locations (Crossnacreevy, C; Downpatrick, D; Limavady, L), which had given rise to differences in wheat specific weight (SW), to examine the relationships between apparent metabolisable energy (AME) concentration, broiler performance and wheat SW. 2. The diets contained (g/kg): wheat 744, casein 142, blended vegetable fat 50, dicalcium phosphate 22, potassium bicarbonate 10.8, sodium bicarbonate 7.5, arginine 5, methionine 2, binder 8, trace minerals/vitamins 7.2, titanium dioxide 1.5. The diets were heat-treated (80°C for 2 min) prior to pelleting (3 mm die). 3. SW ranged from 63 to 77 kg/hectolitre (hl), averaging 66, 69 and 76 kg/hl at D, C and L, respectively. In vitro viscosity of the wheat samples ranged from 5.2 to 17.5 cps and thousand grain weight (TGW) from 33.4 to 47.3 g. Mean TGW was similar at C and D (38.7, 37.0 g) but higher at L (43.1 g). In vitro viscosity was similar at C and L (11.2, 10.2 cps) but somewhat higher at D (14.4 cps). Crude protein (6.25 N) ranged from 116 to 147 g/kg and tended to be higher at D. Starch, which ranged from 612 to 656 g/kg, was least at D (617 g/kg) and greatest at L (641 g/kg). 4. Crude protein, crude fibre and total non-starch polysaccharide (NSP) were negatively correlated with SW, the R 2 being respectively 0.38 (P <0.05), 0.16 (NS) and 0.45 (P <0.05). TGW and starch concentration were positively correlated with SW (R 2=0.70, 0.44, respectively). There was a weak (NS) negative relationship (R 2=0.19) between in vitro viscosity and SW. For both TGW and in vitro viscosity, correlations improved when variety was taken into account (R 2=0.95, 0.92, respectively). 5. There were no significant effects of variety on dry matter (DM) intake or live weight gain (LWG). Gain: food was significantly higher (P <0.05) for Consort than for the other three varieties and the metabolisable energy ratio (ME:GE) just failed to attain significance (P =0.062). Calculated wheat AME (MJ/kg DM) was significantly (P <0.05) higher for Consort than for the other three wheats. There was a good correlation (R 2=0.49) for the total data set between gain:food and ME:GE. In vivo viscosity varied from 13.6 to 28.6 cps for individual treatments and was significantly affected by variety (P <0.001). 6. Although there were no significant differences in DM intake or LWG due to site the values for L (SW 76) were 6 and 5% lower, respectively, than for D (SW 66). Gain:food was lower (P <0.05) for C (SW 69) than for D. ME:GE, wheat AME and ME:gain were not significantly different between sites. 7. There was a weak (R 2=0.18) positive relationship between ME:GE and SW corresponding to a 2.5% increase in energy value for a 10 kg/hl increase in SW and no relationship between gain:food and SW. When variety was taken into the regression the slope was similar but R 2 increased to 0.82. 8. ME:GE and wheat AME concentration were negatively correlated with wheat in vitro viscosity (R 2=0.64, 0.55, respectively). 9. It was concluded that in vitro viscosity appears to provide a better basis than SW for prediction of the nutritive value of wheats of unknown variety. If the variety is known then SW could be used to predict energy value. However, the effect of quite a large change in SW (10 kg/hl) was relatively